We review the first six years of radio observations of Supernova 1987A. The evolution can be divided into two phases: the initial radio outburst which lasted a few weeks, and the period from mid-1990 to the present, during which the radio emission has steadily increased. Both phases can be explained by a small fraction (0.1–0.5%) of the post-shock thermal energy being converted to energy in relativistic particles and magnetic fields, which give rise to synchrotron radiation. The optical depths, densities and density profiles for the pre-shocked circumstellar material are somewhat different for the two phases, but consistent with models of the density structure of the material within the circumstellar ring. New high-resolution radio observations show that the SN shock front is already at about three-quarters of the radius of the circumstellar ring, and that there exists a bright equatorial component of emission aligned with this ring which is probably due to a polar density gradient in the ‘hourglass’ structure.

Introduction

Radio studies of supernovae began with the detection of SN 1970G in M101 (Gottesman et al. 1972; Allen et al. 1976), though it wasn't for another decade that detailed radio light curves were available for a statistically useful sample of supernovae. Mainly through the work of Weiler, Sramek and collaborators (this volume) at the Very Large Array, there are now over a dozen well-studied examples of radio supernovae (RSN).

The bright O I λ11287 line observed in SN1987A is produced by the Bowen fluorescence with Lyβ and comes from regions that lie within a Sobolev length (δR ∼ 10−3RSN, the maximum distance over which fluorescence can work) from hydrogen rich gas ionized by the 56Co decay. Its strength relative to hydrogen lines (e.g. Brγ) depends on the O/H relative abundance in the ‘fluorescent region’ and on the density (i.e. the filling factor) of the gas. The observed evolution of λ11287 can be successfully understood using a relatively simple theory which takes into account the effects of transfer in the O I lines and is the generalization of the classical theory of Bowen fluorescence.

The most important result is that the time evolution of the relative intensities and profiles of O Iλ11287 and Brγ is a powerful diagnostic to determine:

1. – The filling factor of the hydrogen rich gas;

2. – The pre-SN O/H relative abundance;

3. – The amount of small scale mixing between hydrogen and oxygen rich regions and its radial stratification.

In SN1987A the results are the following:

1. –Inside 2000 km/s the hydrogen rich material is clumped with f ≃ 0.1

2. – Outside 2000 km/s the gas has f ≃ 1 and the oxygen relative abundance is quite low: O/H≃ 5 × 10−5, indicating that only the pre-SN oxygen is fluorescently coupled with hydrogen.

3. […]

Freeze out effects and the IR-catastrophe are discussed for SN 1987A and for Type Ia SNe. We show that the light curves of the optical lines in SN 1987A provide strong evidence for the IR-catastrophe. We also argue that most optical lines are dominated by non-thermal excitation after ∼ 800 days. The level of this emission is set mainly by the total mass of the elements. Models of the [OI]λλ6300 – 64 light curve show that an oxygen mass of ∼ 1.5M⊙ is needed. Light curve models for Type Ia SNe display a sharp decrease in the optical flux as a result of the IR-catastrophe at ∼ 500 days, producing UBV-photometry inconsistent with observations of SN 1972E by Kirshner & Oke (1975).

Introduction

Observations of SN 1987A, but also a number of other Type II and Type Ia SNe, at late stages have made it possible to study a number of new features in the evolution of the SN ejecta from explosion to the remnant stage. Here we discuss some recent results in this evolution. A more complete review of the background physics can be found in Fransson (1993).

SN 1987A

It is now well established from the bolometric light curve that ∼ 0.07 M⊙ of 56Ni was created in SN 1987A, and that this is responsible for most of the observed emission from the SN during the first ∼ 800 days. Being based on the bolometric light curve, this is a fairly model independent conclusion.

The nebular spectra of supernovae differ from those of better-known emission nebulae in that many of the emission lines are optically thick. Here we sketch the theory for interpreting such spectra, and show how it can be used to interpret prominent emission line systems in the spectrum of SN 1987A. As examples, we describe: (1) a simple method to infer the density of O I from observations of the evolution of the doublet ratio in [OI]λλ6300; (2) new kind of hydrogen recombination line spectrum; (3) an analysis showing that the Ca II infrared emission lines must come from primordial, not newly-synthesized, calcium; (4) a theory for the Fe/Co/Ni emission lines that shows that the inner envelope of SN 1987A must have a foamy texture, in which low density radioactive bubbles of Fe/Co/Ni reside in a massive substrate of hydrogen, helium, and other elements.

Introduction

Conventional wisdom holds that supernova explosions produce most of the heavy elements in the universe, and a major goal of astronomy is to test this hypothesis through observations of supernova spectra. For this purpose, SN 1987A should be a Rosetta Stone. We have observed its spectrum in far greater detail than that of any other supernova: at wavelength bands, such as gamma rays and far infrared, where no other supernova has been observed; with almost daily (nightly!) observations continuing for more than seven years after outburst; and with unprecedented spectral resolution (McCray 1993).

We perform hydrodynamical calculations of the collision between the supernova ejecta and circumstellar matter for SN 1987A and SN 1993J. For SN 1987A we predict light curves of X-ray emissions from the shocked ring. For SN 1993J, thermal X-rays from the shocked circumstellar matter can consistently account for the observations with ROSAT, ASCA, and OSSE.

Introduction

The supernova ejecta collides with the circumstellar matter (CSM) if its progenitor was undergoing significant mass loss. Shock waves arising from this collision compress and heat the ejecta and the CSM. The emission from the shocked material strongly depends on the density distributions of the ejecta and the CSM, thereby providing important information about the nature of the CSM.

SN 1987A

The images from the European Southern Observatory (ESO) (Wampler et al. 1990) and the Hubble Space Telescope (HST) (Jakobsen et al. 1991) revealed the presence of a ring-like structure at ∼ 6 × 1017 cm from SN 1987A. The outermost part of the supernova ejecta is expanding at ∼ 104 km s−1 (Shigeyama & Nomoto 1990), thus being expected to collide with the ring at ∼ 10 years after the explosion.

Hydrodynamical model

The progenitor of SN 1987A had once become a red supergiant (RSG) and then contracted to a blue supergiant (BSG) before the explosion (for reviews, see Arnett et al. 1989, Hillebrandt & Höflich 1989, Podsiadlowski 1992, and Nomoto et al. 1993a).

Theoretical light curves and spectra of X-rays and γ-rays from SN 1987A are calculated by the Monte Carlo method, based on a model built up from the early observations of neutrinos and optical light. Comparison of the predicted radiation with observational results obtained later confirms the radiation mechanism of supernovae: γ-rays are emitted in the decays of radioactive 56Co and X-rays are generated by the Compton degradation of these γ-rays. It also suggests that large scale mixing occurred and clumpy structure was formed inside the ejecta. These findings lead us to construct the model with a new distribution of elements, which is determined through comparisons of observations of X-rays and γ-rays with numerical simulations based on the assumed distribution. Using this model, the subsequent X-ray and γ-ray emission is predicted: the light curves of X-rays and γ-rays as well as their spectral evolution are in very good agreement with that expected from the radioactive decays of 56Co and 57Co. The mass of newly synthesized 44Ti and the emission from the neutron star will be determined by future satellite and balloon-borne observations.

Introduction

SN 1987A has given us an invaluable chance to examine supernova theory, which has predicted the emergence of X-ray and γ-ray radiation from supernovae. Several possible mechanisms for the X-ray and γ-ray emission have been discussed, such as collision of the ejecta with circumstellar matter, nonthermal radiation from a pulsar, and Compton degradation of the line γ-rays emitted by radioactive nuclei.

X-ray spectroscopy can provide vital information about the progenitors and environments of supernova remnants. Plasma diagnostics and spectral modelling can be used to infer the energy of the remnant, the density and composition of the surrounding medium, and the degree of equilibrium in the shock heated gas. A new generation of X-ray spectrometers, the first of which was the Broad-Band X-Ray Telescope (BBXRT), has improved our ability to make precise measurements of X-ray line fluxes and energies. We summarize the results obtained from the BBXRT mission. These include a definitive measurement of the Fe K line centroid in the Tycho remnant, production of the first narrow-band X-ray maps (of Puppis A) and the first measurement of an electron-ion equipartition timescales in evolved remnants.

Introduction

Supernova remnants may be grouped into three broad categories, based on their X-ray and radio morphologies. The first of these shows shell-like structure in both bands. The X-rays from these are thermal, arising from the shock heating of ejecta and interstellar material. Prominent examples of this class of remnant are Tycho and the Cygnus Loop. The second category shows centrally peaked emission in both bands; these are the plerions, or Crab-like remnants, after the class archetype. The X-ray emission is a non-thermal power law, dominated by synchrotron processes from the energetic electrons produced by the pulsar. A third category combines elements of the previous two.

Introduction

Several multidimensional computations of hydrodynamics related to supernovae have been completed, and are summarized here. More detail may be found in Arnett 1994a,b, Arnett & Livne 1994a,b, and Livne & Arnett 1993. The hydro code PROMETHEUS is based upon an implementation of the piecewise-parabolic method (PPM) of Colella & Woodward 1984, as described in Fryxell et al. 1991. A detailed comparison of PPM with other schemes is given in Woodward & Colella 1984. The method constructs the physics of the flow between grid points by a nonlinear solution of the equations of continuity of mass, momentum and energy (the Riemann problem) rather than the usual mathematical approach of a Taylor expansion about the grid points. This gives it better resolution per grid point, which is highly desirable for multidimensional problems. Although the effort required per grid point is greater, the number of such points is less (often much less) for a given level of accuracy. Because the computational load per grid point is greater, more realistic physics (reactions, radiation, gravity, etc.) may be added before affecting the runtime significantly. Thus PPM is well suited for multidimensional problems with significant physics beyond the bare hydrodynamics.

The Prometheus project was an effort by the author, Bruce Fryxell and Ewald Müller, to implement a “state of the art” hydrodynamic method with realistic microphysics for stellar problems. After an extensive study of several methods, and direct comparison of the resulting codes (Fryxell, Müller & Arnett 1989), PPM was chosen as the preferred method.

We review critical physics affecting the observational characteristics of those supernovae that occur in massive stars. Particular emphasis is given to 1) how mass loss, either to a binary companion or by a radiatively driven wind, affects the type and light curve of the supernova, and 2) the interaction of the outgoing supernova shock with regions of increasing ρr3 in the stellar mantle. One conclusion is that Type II-L supernovae may occur in mass exchanging binaries very similar to the one that produced SN 1993J, but with slightly larger initial separations and residual hydrogen envelopes (∼1 M⊙ and radius ∼ several AU). The shock interaction, on the other hand, has important implications for the formation of black holes in explosions that are, near peak light, observationally indistinguishable from ordinary Type II-p and Ib supernovae.

Some Generalities

There is broad agreement regarding the qualitative evolution of single stars sufficiently massive to ignite carbon burning non-degenerately (e.g., Woosley & Weaver 1986; Weaver & Woosley 1993; Nomoto & Hashimoto 1986, 1988). Given the usual, relevant caveats about the treatment of convective mixing, convective overshoot, and semiconvection, it is agreed that stars of approximately 8 to 12 M⊙(±1 M⊙ depending upon initial helium abundance and convective parameters) will not proceed to silicon burning in hydrostatic equilibrium, but will stop prior to central neon ignition and experience a complicated subsequent evolution in which degenerate flashes play an important role.

Cosmology is a general relativistic topic. General relativity must replace Newtonian mechanics for systems in which the mass and spatial dimensions have similar orders of magnitude, M ~ r. If the density of the Universe were a constant, M/r ≈ ρr3 /r ≈ ρr2, and for large enough length scales, the condition M ≈ r would eventually be met. Let us take the current density of 10-29g cm-3 (= 7.4 x 10-58cm-2 in geometrized units). Then, for length scales of r ≈ 3.7 x 1028cm (≈ 104 Mpc), calculations must be carried out using general relativity. This length is of the order of magnitude of the observable Universe, and corresponds to objects about 1010, years old, near the commonly accepted age of the Universe.

The cosmological principle

The Newtonian interpretation of the cosmological principle must now be reconsidered. It stated that at a given time all thermodynamic parameters are homogeneously and isotropically distributed. But what kind of time? Now every observer has his proper time and thus the time that the cosmological principle takes as a reference must be specified without invoking privileged observers or systems with peculiar characteristics. General relativity assures us that at any point there is a free-falling observer, for whom nature is explained in terms of the laws of special relativity. Observers in free-fall are called fundamental observers in cosmology and the particle of cosmic fluid that they ride is called a fundamental particle, or it could be called a fundamental galaxy, taking galaxies as the pieces from which the Universe is built.

Main properties

The fluid of stars in a galaxy closely resembles the fluid of molecules. The fluid of galaxies does not. Let us examine three main differences.

(a) The distribution of galaxies is not chaotic. There are groups, clusters, superclusters, and there is a large-scale structure, which we are now beginning to realize. The distinction between clusters and superclusters is not sharp, and as superclusters are elements of larger structures their limits and sizes are difficult to establish. The largest observed structures are as large as the limits of the deepest surveys. A continuous spectrum of inhomogeneities describes structures larger than a galaxy better than it describes discrete objects such as stars and galaxies.

The relative increase in the density with respect to the mean density δ is about 102 –103 for clusters. Typical intercluster distances of about 5 Mpc and inter-supercluster distances of about 25 Mpc are revealed by cross-correlation studies. The value of δ for a galaxy is about 105.

The large-scale structure of the Universe is complex. Many clusters are aligned in huge filaments. Others seem to form sheets. There are also voids which are apparently deplete of galaxies. A simplified picture of the large-scale structure might consist of an ensemble of large polyhedral voids. In the limiting sheets separating two adjacent voids there are clusters. Along the limiting line intersections there are more clusters. At the limiting vertices there are still more clusters. The linear dimensions of the voids are typically 20 – 50 Mpc.

This small book is intended as a general introduction to astrophysical fluid dynamics. The reader is presumed to possess a knowledge of basic physics, namely, classical physics, elements of relativity, and introductory ideas about quantum mechanics. No previous knowledge of fluid dynamics or of astrophysics is required, these topics being introduced in the book. Although fluid dynamics may constitute a complementary, original, natural, fecund, unexplored, simple, and enjoyable way to introduce astrophysics, the topic of astrophysical fluid dynamics is a promising, distinct, and particularly wide branch of astrophysics at the present time.

The first part of the book (Chapters 1–4) deals with basic fluid dynamics. Although it could also be used for non-astrophysical purposes, it was written with the former in mind. It often includes cosmic examples that are mainly related to a stationary, static, and stratified atmosphere. These conditions provide the greatest simplification while maintaining a high degree of astrophysical interest.

Following the first chapter on classical fluids, Chapter 2 is devoted to relativistic fluids. The early introduction of relativistic fluids is necessary, as many cosmic fluids, and the cosmic fluid itself, are relativistic. One important advantage is that radiative transfer can be developed as transport in a relativistic fluid, thereby avoiding the usual classical mis-interpretation of the radiative Boltzmann equation. Plasmas and magnetohydrodynamics are also included because of their growing interest in the field of astrophysics. The important role played by magnetic fields in a large sample of cosmic systems is only now beginning to be appreciated.